The aim of the project is to build a tool to assist journalists by tracking article analytics and providing helpful information based on the article performance.

The tool is being implemented as a Slackbot and uses integration with Google Analytics to track article performance.

The user flow is as follows:

1. A user adds the slackbot to their Slack account using the Slack Button and by completing the OAuth integration.

2. The user then grants the application access to Google Analytics using Google's OAuth integration.

3. The user selects the appropriate Google Analytics web property.

4. Once the Slack and Google integrations are setup the /add slack slash command will be available. A new article can be added using the slash command.

5. Once the article is confirmed, the application will track the article using the Google Analytics Realtime API

6. The slackbot will send the user a message if the analytics show a large increase or decrease in article traffic.

The app is built using:

• Elixir 1.3
• Phoenix 1.2
• Node.js 5.3.0
• PostgreSQL 9.4
• Redis 3.2

Homebrew is a package manger for OSX, we'll use this install node and elixir.

ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

Install Elixir via Homebrew

brew install elixir

Install PostgreSQL using the Postgres.app

Install noedejs via Homebrew

brew install nodejs

Install redis via Homebrew

brew install redis

• Clone the repo
• Copy the dev.secret.exs.sample file to dev.secret.ext and add the appropriate values
• Install dependencies with mix deps.get
  • If you get errors here, open a new terminal window. Some commands don't yet have proper binding if you've just installed elixir from Homebrew.
• Create and migrate your database with mix ecto.create && mix ecto.migrate
• Install Node.js dependencies with npm install

To start the app locally:

We're using Redis to queue and schedule background jobs such as refreshing OAuth tokens and regularly pulling data from Google Analytics.

Start redis by running: redis-server

We're using Phoenix as the webserver for the application - this serves the websites, manages websocket connections for the slackbots and schedules and runs the background jobs.

Start Phoenix endpoint by running: mix phoenix.server

Now you can visit localhost:4000 from your browser.

The project is hosted on Heroku and available at https://espi-dni.herokuapp.com. Deployment can be done by pushing to the Heroku git repository.

We are using Codeship for continuous integration. Codeship will run tests for any pushes to any branch for the project on github.

Deployment to production is automatically carried out as part of the codeship build process. When changes are pushed to master, all tests are run, and if the build passes, the current state of master is deployed to heroku (https://espi-dni.herokuapp.com. As part of the deployment process, any pending migrations will be run, and the server will be restarted.

The application requires API integration with both the Google Analytics API and Slack API.

To use any of the Google Analytics APIs a Google Client ID and Client Secret are required. In production these are added as environment variables GOOGLE_CLIENT_ID and GOOGLE_CLIENT_SECRET. In development the keys can be added directly to the dev.secret.exs file which is is not added to git.

The application is using Google Analytics Realtime API. This API is currently in limited beta and access must be requested to use the API.

The application is using Google's Oauth2 API to retrieve Google Oauth2 access tokens to view Google Analytics on a users behalf.

To use any of the Slack APIs a Slack Client ID and Client Secret are required. In production these are added as environment variables SLACK_CLIENT_ID and SLACK_CLIENT_SECRET. In development the keys can be added directly to the dev.secret.exs file which is is not added to git. This Slack ClientID and Secret are obtained in the Slack App Management Console.

Any requests that are received from slack must contain a token from slack verifying the request is actually coming from slack. The token is stored as an environment variable SLACK_TOKEN on production and in dev.secrets.exs on development. The token can be obtained in the Slack App Management Console.

The slackbot connects to a slack team using the Slack Realtime API. This opens a real-time web socket connection that is managed by the BotSupervisor module

Spike notifications are sent via the Slack Web API.

To work on a local version of the project, it makes sense to create a personal slackbot that can send requests to your local environment using a tool like ngrok.

• Before you start, install ngrok (or a similar tool) and take note of the public URL your application is running at e.g. https://foobar.eu.ngrok.io
• Visit https://api.slack.com/apps and hit 'Create New App'
• Give the app a name (let's say post-dev) and pick the team you want to use it with.
• Next click on 'Interactive Messages' and click 'Enable Interactive Messages'.
• Go to 'Slack Commands' and create a new commad, let's call it add-dev
• Set the Request URL for the slash command to your local ngrok /slack/articles/new path. e.g. https://foobar.eu.ngrok.io/slack/articles/new
• Add a bot user to your application, give it name, (e.g. 'post-dev')
• Set the Request URL for interactive messages to the /slack/messages/new path of your local environment. e.g. https://foobar.eu.ngrok.io/slack/messages/new
• Save the Client ID, Client Secret and Verification Token to your local environment and restart your application.
• Add another method in your SlackArticleController module to respond to the new add-dev slash command you added

e.g. Add a second new method to match the slash command locally (or just change the existing one locally only)

  def new(conn, %{"command" => "/add-dev", "text" => url}) do
    changeset = Article.changeset(%Article{}, %{url: url, user_id: conn.assigns.current_user.id})
    if changeset.valid? do
      render(conn, "confirm.json", url: url)
    else
      text conn, error_string(changeset)
    end
  end

You should now be able to use your newly created slackbot with your local environment. Note, if your ngrok endpoint changes, you will need to update the endpoints you've set in your slackbot for the slash command and interactive messages